The Axia Guide to Choosing Category Cable


by Stephen H. Lampen
Multimedia Technology Manager
Belden Electronics Division, Belden CDT

If you've gotten to this paper, you've probably read the promotional literature for the Axia line of products, and maybe the "Introduction to Livewire". If you have, then you know that this is a new type of audio control product, one that is an amalgam between computer networking and audio processing. It is a 100Base-T network running audio as the data stream, with all the advantages of cost, simplicity, power and elegance of a computer network.

But it also means that you will be installing a network based on network cables, usually called Category cables. If you're an old broadcast engineer, like me, this move to a networked architecture, and these strange unshielded cables, may be a leap of faith. But let me assure you, many other engineers have taken this leap and not only survived but prospered.

There are lots of other people (those IT network types) on the other side. And the Telos/Axia folks are more than eager to help you through the learning process. Because that's really all you have to do, learn about this "new way" of running audio. There are tens of thousands of very reliable data networks running Ethernet(R) around the world, and have been for many years. Hey, it's not lonely at all. In fact, it's kind of crowded! Now you probably didn't become a broadcast engineer or installer to end up a data dude (or gal), but, sorry, you are now officially a network installer.

And one of the things you may have to learn about is cable, specifically "premise/data" cable that comes in a number of flavors, called Categories. Many people call these "Cat" cables, short for Category. We'll discuss Cat 5e, and Cat 6 (such as Figure 1), a bonded-pair Category 5e, Belden 1700A. We'll examine how Category cables are different, and how they are the same, how to choose good cable from bad cable, and how to get the most bang for the buck.

A (Very) Short History Of Computer Cables

When computers were first invented, they were running at lightning-fast speeds like 1 megabit per second. (That was fast for the 1950s.) No twisted-pair cable could carry a signal like that so computers ran on coaxial cable. Computer designers looked longingly at twisted pairs for one reason: noise rejection. Twisted-pairs can be run as "balanced lines" that dramatically reduce electromagnetic noise picked up by a cable.

Where does this noise come from? Everywhere! Motors, generators, fluorescent light ballasts, lighting dimmers, even computers, like the one you're probably looking at right now, are all sources of electromagnetic noise. Two-way radios, medical machinery, and, of course, radio and TV broadcast transmitters are wonderful sources. And then we have broadband sources, like the sun, an excellent source of electromagnetic noise, which is why things get electrically "quiet" at night and you can hear that AM radio station a thousand miles away.

How Twisted Pairs Work

To understand how twisted pairs work, we have to understand "balanced lines". To understand balanced lines, we start with Figure 2.

Figure 2 shows a battery and a light bulb. We get the electricity to flow through the light bulb by attaching two conducting paths, usually wires. I've put two arrows to show how the electricity will flow, out of the negative terminal of the battery, through the light bulb, and back to the positive terminal.

This often confuses readers. Why does the electricity move in opposite directions? Because it is a "circle" of conducting pathway. It's like a race track. If you're in a race and you look across to the other side of the track, those cars are going in the opposite direction.

It doesn't matter how close These wires are together, like Figure 3, or even if we twist them together like Figure 4. Nothing has changed. It's still a circle of electricity, which is why we give it the Latin name for circle: circuit.

Now that we have a twisted pair, as in Figure 4, we have to cover the wires with a non-conductor, like plastic, so that they can't touch each other. If they did, electricity takes the path of least resistance where they touch, instead of the light bulb that has a lot of resistance. Our circuit won't be as long as it's supposed to be. It would be a short circuit!

If we replaced the battery and light bulb with something else that produces electrical signals and something at the other end that uses them, amazingly little has changed, such as Figure 5.

In this example we have a microphone as the source of electricity. A microphone converts acoustical energy into electrical energy. At the other end we have a speaker that turns electrical energy into acoustical energy, i.e. sound. (Of course, we would have to have a preamplifier and power amplifier inside the speaker, for all you nit-pickers! That's why I put knobs on the speaker!)

If we put the microphone in a piano and hit "Middle A" (440 vibrations per second) then the diaphragm in the microphone would move in and out 440 times per second, the two arrows on our twisted pair would reverse direction 440 timers per second, and the speaker cone would move in and out 440 times per second. We would hear that note: 440 Hertz. If this were a perfect microphone, perfect cable, and perfect speaker, we would hear the note as if our ear were in the piano where the microphone is.

Twisted pairs, when connected as a balanced line, reduce noise. And the secret is a device at each end of the pair called a "transformer". Figure 6 shows a twisted pair with a transformer at each end. Transformers are just coils of insulated wire, but they can pass signals between them.

You'll notice noise (big yellow arrow) coming from outside. The insulation on the wire can't stop it, so when the noise hits each wire it creates ("induces") a signal on the wire (those little yellow arrows). Since there are two wires, there are two noise signals. Those

noise signals travel to the end of the cable where they meet each other inside the transformer and cancel each other out.

Fancy Words

The noise signals are moving in the same ("common") direction, so they are called "common-mode noise". If we measure them, we're interested in how well the transformers cancel out the noise, compared to the noise that might get through and not get cancelled out. So this measurement is a ratio of the stuff that gets through to the rejected noise signal, called "common mode rejection ratio" or just CMRR, for short.

You might recall that the signal we want to travel down the cable is traveling in opposite directions (as in Figure 6). Since those arrows are moving in different directions, we call this a "differential" signal. And the key is that, if we could measure the signal on the two wires and mathematically add them together, the total should always equal zero, since they're opposite signals.

If the differential signal doesn't equal zero, it means that the twisted pair is not balanced and any noise on the cable would not be completely cancelled out and that noise would be added to the desired signal.

Once noise is added, it is very difficult to get rid of. It's much easier to get rid of it before it gets included with the signal.

An Ideal Twisted Pair

Table 1 shows basic factors that go into making, and measuring, a twisted pair. We have a saying in cable manufacturing: "Physicals equal Electricals". That means, anything you do that physically changes the cable will also change its electrical performance.

Requirement Variations In Cable Parameter Measured In
Spacing Impedance Return Loss Decibels (dB)
Capacitance Capacitance Unbalance Picofarads (pF)
Length Resistance Resistance Unbalance Ohms (Ω)
Size Resistance Resistance Unbalance Ohms (Ω)
Timing Phase Degrees (°)

Table 1

For instance, if the two wires in our twisted pair are not the same length, then the noise signals will not arrive at the same time. They will be "out of phase" and the cancellation will not be perfect. Also one wire being longer, will have more resistance, so the two noise signals will not have the same intensity and therefore not cancel out.

The same applies to size (wire gage, AWG). Even if they are slightly different in size, it can have a big effect on noise rejection. We know how long the cable is, and we know what the resistance per foot (or per meter) should be, so we can easily calculate what the value should be. If it's different than that theoretical value, then we have an unbalance, one wire is more or less resistance than the other, and will allow some percentage of noise to be passed to the next device.

The two wires also need to be close together. Notice in Figure 6 how the two wires are spread apart. Now you might understand that this is a very poor twisted pair. And the reason is that the noise signal hits one wire and, a tiny fraction of a second later, hits the second wire. This bit of time means that the two noise signals will not arrive at the same time at the transformer. They will be out of phase and not completely cancel out.

The distance between the two wires also affects the "impedance" of the cable. The impedance is a combination of capacitance, inductance, and resistance, everything that tries to "impede" signal flow. At low frequencies, below 1 MHz, the cable isn't long enough for the impedance to make a difference so the impedance at those frequencies is often ignored.

Above 1 MHz, impedance becomes more and more important. At 100 MHz (like Cat 5) it is very important. Cables that are not the correct impedance (100 ohms for Category cables) will reflect part of the data signal back to the source. This is called "return loss" and is a good way to compare cables. It is especially interesting to note if this is "typical" return loss (can be higher or lower), or "maximum" return loss (no worse than). Just having a maximum return loss number is a good indication of a superior Category cable.

Now you might begin to understand what the definition of a perfect balanced line is. It is a pair of wires, with all passive components attached to them, where each wire is the same impedance in respect to ground. In other words, two wires that are electrically identical.

Perfect Pairs

Of course, there's no perfect anything, but manufacturers have ways of getting closer to perfection. One of these ways is to make your own bare wires. Many cable manufacturers buy their bare wire from a "wire mill". If they buy a 24 AWG (gage) wire, how will they know it's perfect? (They don't.) And the wire mill probably doesn't know what this wire will be used for. It could be for door bell wire, where almost anything would work just fine. Variations in size may already be in the wire when it is bought. But if a manufacturer has its own wire mill, like Belden, then the wire for Category cable will have much greater precision than the same gage wire used in other applications. And that precision is crucial to good performance.

One way of keeping a twisted pair close to each other, is just that: twist them together. That may sound easy but it is not. If the tension on one wire is slightly more (or slightly less) than the other wire, the two wires will be different lengths, one wire will be closer to straight with the other wire wound around it, bad news for data cable. This pair will suffer from resistance unbalance and timing problems, as listed in Table 1.

And twisted pairs in a finished cable, when bent and flexed while installed, cause the pair to "open up" changing the impedance, increasing the return loss, changing capacitance, and allowing noise to get in.

Bonded Pairs

One solution for this problem is bonded pairs. This technique sticks the two wires together as they are twisted. This dramatically improves impedance stability, reduces capacitance unbalance, and reduces return loss. The only bad thing is that you have to split the wires apart when you install them, which adds a few seconds to the installation of each connector.

That minor amount of time is more than offset by the fact that bonded-pair cables retain their performance specs after they are installed. And this means fewer call-backs to fix bad cables. Figure 7 shows bonded-pair Belden 1872A MediaTwist Category 6.

Base What?

Axia uses 100Base-T and 1000Base-T signals. 100Base-T means 100 Megabits of data per second ("100 Mbps") based on twisted pairs ("T"). So you can probably guess that 1000Base-T means 1000 Megabits (1 Gigabit) of data (Gbps) on twisted pairs. Most Axia devices run on 100Base-T, although much of the hardware can also handle 1000Base-T.

Standards

Category cable standards are set by a joint committee of the Telecommunication Industry Association (TIA) and the Electronic Industry Association (EIA). The committee is called TIA/EIA 568. Their current set of standards is called TIA/EIA 568-B.2

Table 2 and 3 show the standards for Category 5e and Category 6. Some of the Belden products that are made to this standard are in the last column. The 568-B.2 standard contains specifications for Category 3 (now used as telephone cable), Category 5e, and Category 6. They have dropped Category 4 and Category 5 from the standard, so it's very hard to buy that type of cable. They're working on the next standard, 10GBase-T, 10 gigabits on four-pair UTP.

TIA/EIA 568A Partial Specifications for Category 5e per 100 meters (328 feet)
Frequency Minimum PSNEXT Maximum Attenuation Minimum ELFEXT Return Loss Minimum PSACR Delay Skew Belden Products
.772 MHz 64 dB 1.8 dB 60.8 dB --- --- 45 nanoseconds 1583A
1585A
1500A
1501A
1700A
1701A
7988R
7988P
1 MHz 62.3 dB 2.0 dB 48.7 dB -20 dB 60.3
4 MHz 53.3 dB 4.1 dB 42.7 dB -23 dB 49.2
8 MHz 48.8 dB 5.8 dB 40.8 dB -24.5 dB 43
10 MHz 47.3 dB 6.5 dB 36.7 dB 25 dB 40.8
16 MHz 44.3 dB 8.2 dB 34.7 dB 25 dB 36
20 MHz 42.8 dB 9.3 dB 32.8 dB 24.3 dB 33.5
31.25 MHz 39.9 dB 11.7 dB 30.9 dB 23.6 dB 28.2
62.5 MHz 35.4 dB 17 dB 24.8 dB 21.5 dB 18.4
100 MHz 32.3 dB 22 dB 20.8 dB 20.1 dB 10.3

Table 2

TIA/EIA 568A Partial Specifications for Category 6 per 100 meters (328 feet)
Frequency Minimum PSNEXT Maximum Attenuation Minimum ELFEXT Return Loss Minimum PSACR Delay Skew Belden Products
.772 MHz 74 dB 1.8 dB --- --- --- 45 nanoseconds 7881A
7882A
1872A
1874A
7851A
7852A
7989R
7989P
1 MHz 72.3 dB 2 dB 64.8 dB 20 dB 70.3
4 MHz 63.3 dB 3.8 dB 52.7 dB 23 dB 59.5
8 MHz 58.8 dB 5.3 dB 46.7 dB 24.5 dB 53.4
10 MHz 57.3 dB 6 dB 44.8 dB 25 dB 51.4
16 MHz 54.3 dB 7.6 dB 40.7 dB --- 46.7
20 MHz 52.8 dB 8.5 dB 38.7 dB 25 dB 44.3
25 MHz --- --- 36.8 dB 24.3 dB ---
31.25 MHz 49.9 dB 10.7 dB 34.9 dB 23.6 dB 39.2
62.5 MHz 45.4 dB 15.4 dB 28.8 dB 21.5 dB 30
100 MHz 42.3 dB 19.8 dB 24.8 dB 20.1 dB 24.3
125 MHz 40.9 dB 22.4 dB 22.8 dB --- 18.5
150 MHz --- --- 21.2 dB 18.9 dB ---
155 MHz 39.5 dB 25.2 dB 20.9 dB 18.8 dB 14.3
160 MHz --- --- --- 18.7 dB ---
175 MHz 38.7 dB 26.9 dB 19.9 dB --- 18.8
200 MHz 37.8 dB 29 dB 18.7 dB 18 dB 8.8
225 MHz 37 dB 31 dB 17.7 dB --- 6.1
250 MHz 36.3 dB 32.8 dB 16.8 dB 17.3 dB 3.5
300 MHz --- 36.4 dB --- --- ---

Table 3

Be aware that Table 2 and Table 3 are the "standards", the minimum (or maximum) that all Cat 5e or Cat 6 must meet. The Belden cables listed at the end easily meet, and very often exceed, these requirements.

What They Mean

The specifications in Tables 2 and 3 are written in the acronym language of the data world. Here's what those acronyms actually mean, along with some other terms used to describe data cables.

NEXT means "near-end crosstalk". At the near end (source end), the transmitted signal is the strongest. Transmitting pairs can interfere with other signals on other pairs. This is called "crosstalk". The 568B.2 standard specifies the minimum crosstalk value at various frequencies.

PSNEXT is "power sum near-end crosstalk" that looks at the effect of all adjacent pairs to the one under test rather than just pair-to-pair. Such testing is then the "worst case" where all pairs are energized, such as in 1000Base-T. Results of each combination are averaged together.

ATTENUATION is signal loss and is common to all signal carrying systems. Attenuation is measured in decibels (dB). Decibels are logarithmic. Data signals -40 dB down (one-ten thousandth of the original intensity) are fully and easily recovered. This is not surprising to most audio/video engineers considering that analog microphone signals are often –60 dB.

FEXT is "far-end crosstalk". The far end of the cable is where the signals are weakest, where attenuation has already reduced the signal level so that crosstalk can have a significant effect

ELFEXT is "equal level crosstalk". If you're interested in just the crosstalk numbers, then you would subtract the attenuation from FEXT. What is left is crosstalk, all at the same level ("equal level")

ACR is "attenuation-to-crosstalk ratio". ACR subtracts the crosstalk from the attenuation, to indicate the overall performance of a cable. Positive ACR, especially at high frequencies, can be an indicator of superior cable performance. ACR is very similar to "signal-to-noise ratio" in the analog audio/video world. Therefore, ACR can be valuable where multiple data signals travel down a four-pair cable, such as 1000Base-T networking.

PSACR is "power-sum attenuation-to-crosstalk ratio" where all pairs are energized around the measured pair and the ACR results averaged. This shows the "signal-to-noise" ratio with everything running, a very good test.

RETURN LOSS shows the variations in impedance within a cable. Impedance variations cause the signal to reflect back to the source, so return loss is the ratio between direct signal and reflected signal. It is measured in decibels (dB). With a larger negative number, more signal reaches its destination, and less of the signal is reflected back to the source. (-30 dB return loss is better than -20 dB return loss.) Return loss is especially effective in showing flaws in cable construction and installation, such as excessive bending or stretching, which affect the impedance of the cable. In link and channel tests, return loss can show the effect of poor, or badly installed, connectors, patch panels, and other passive hardware.

DELAY SKEW is timing differences on a multipair cable. For all cables listed above, the maximum allowed by TIA/EIA 568B.2 is 45 nsec/100m (nanoseconds per 100 meters, 328 feet). Delay skew is especially interesting to designers who are using more than one pair to simultaneously deliver data. Formats such as Gigabit Ethernet(R) ( 1000Base-T) require that the data be split between the four pairs. In such systems, it is essential that the signals arrive at the other end of the cable at the same time.

Note that some Cat 5e and some Cat 6 cables have ultra-low skew. Those low-skew numbers allow you to use these cables for RGB and VGA (and other analog and digital applications that benefit from precision multi-pair delivery) as well as for 100Base-T or 1000Base-T data applications.

Timing variations for a complete system should not exceed 50 nsec (nanoseconds) between any of the four paths. When looking at cable alone, the maximum delay is 45 nsec/100m. This is one of the reasons Belden's MediaTwist(R) (delay skew 25nsec/100m maximum, 12nsec/100m typical) is popular in applications where this is critical.

Be aware that there are "no skew" cables. These have vanishingly low skew, such as 2.2 nsec/100m for Belden 7987R (7987P). But all these cables, and all "no–skew, zero-skew" designs from other manufacturers, accomplish this by having all four pairs with identical twists. While these cables might be good for non-data applications, such as RGB and VGA, they are not Category anything. They won't even pass Cat 3 (telephone cable), and should not be considered for anything requiring true Cat 5e or Cat 6 performance.

On the other hand, there are category cables (Belden 7988R and P, 7989R and P) that are true Cat 5e and Cat 6, where the cable design concentrates on ultra-low delay skew with values of 9nsec/100m (7988) and 10nsec/100m (7989). This allows an installer to use the same cable for data network applications (such as Axia) and also use it for RGB or VGA display.

PAIR TWISTING ("Lay Length"). Tight pair twisting can greatly reduce crosstalk but also has a number of negative effects on the cable. There is more copper used per unit length so the price goes up. And more copper means the signal will take longer to travel down that pair (compared to other pairs) so attenuation and delay skew are worse. It doesn't matter how high your ACR is, or how low your crosstalk is, if you don't have enough signal strength to be recovered at the receiving end! What you truly want is a cable that improves both crosstalk and attenuation to improve ACR, to have positive ACR at a higher frequency, without affecting, or possibly even reducing, delay skew.

IMPEDANCE indicates the ability of a data cable to transfer a signal from one box to another. The impedance of the systems, and boxes it is attached to, specifies the impedance of the cable. The TIA/EIA standard for Category 5e and 6 is 100Ω ± 15Ω (ohms). Some cables meet this spec. Others require the use of a smoothing formula called "Zo-fit". This allows manufacturers to ignore rapid changes in impedance. Belden bonded-pair data cables are tighter than ±15Ω without the Zo-fit function.

BANDWIDTH is the range of frequencies available to be used for signal carrying. It is the "width of the tunnel". However, knowing the width of the tunnel tells you nothing of how the traffic will move through it. This is because data can be compressed to take up less bandwidth.

For instance, a 100 Mbps data signal can fit in a 100 MHz bandwidth. Or the data can be arranged and coded to fit in a 50 MHz bandwidth, a 30 MHz bandwidth, or even smaller. In fact, the 31.25 MHz numbers commonly seen in cable specifications are for a compressed 155 Mbps ("ATM") protocol.

Since the coding scheme is not apparent, only the bandwidth in Megahertz (MHz) can be used to compare potential data handling capacity. You will know the size of the tunnel. Knowing how many cars will fit depends on how you arrange them. If you want to compare the signal-carrying capacity of two cables, compare bandwidths.

Isolation

Another technique to improve performance in found in all Category 6 cables. These work at much higher frequencies (250 MHz) than Cat 5e. And they are intended to carry 1000 Mbps (megabits-per-second), also called 1 Gigabit per second (Gbps).

The crosstalk requirements of Cat 6 are 10 dB harder than Cat 5e, so most manufacturers have solved this problem by putting a divider in the cable, as shown in Figure 8, Belden 7851A "600e" Category 6.

Most dividers are simply an "X" that divides the four pairs into separate quadrants. The dividers in the Belden 7851A, and other cables in that family, are just a little different. It's more a back-to-back "Y" like Figure 9.

In this way, the pairs that are most likely to "talk" to each other are kept as far apart as possible. Another way of keeping the pairs apart to meet Category 6 crosstalk is used in Belden 1872A MediaTwist (Figure 10). MediaTwist spreads the pairs apart, giving each pair a little channel inside the jacket. This is why the cable is "crescent-moon" shaped and not round.

It should be noted that this cable is now over ten years old, ancient for a data cable, and still exceeds Category 6 specifications. In some specs, such as "bend radius" and "pull strength", MediaTwist is still unequalled in the industry.

Like all products, there can be a Chevy, a Jag, or a Ferrari. Figure 11 shows a "Chevy" Category 6, Belden 7881A. Note the divider between the pairs is just a tiny plastic wire. The pairs in this construction are non-bonded. Changes like these make such a cable smaller, lighter, cheaper, and easier to install. You're just trading performance for all those nice things.

Fire Ratings

Fire ratings are defined in the National Electrical Code (NEC). This is a voluntary code, so each city, county, or state may or may not follow this code. It is also interpreted in different ways by Fire Marshals, building inspectors, permit boards and other bodies "having jurisdiction". If your area does not subscribe to the NEC, then you must obtain a copy of their own rules, or at least have someone who can advise you on the local requirements.

Within the NEC, there are different fire ratings, tests that are performed on cables to determine their reaction to a fire. Most often these involve flame spread and smoke production. Most category cables come in two different fire ratings, riser (CMR) and plenum (CMP). There are lower grades (CM, CL2 for example) or higher (LC, limited combustible) that are available. The choice of cable and fire rating is between your architect or system designer and the appropriate legal body having jurisdiction.

Riser rating (CMR) allows cables to be placed vertically between floors without use of a metal conduit. Plenum ratings (CMP) allow cables to be used in drop ceilings or raised floors that are connected to an air conditioning system.

Before you buy the cable for your installation, be sure you have determined which fire rating is appropriate. An inspector can easily require that an entire wiring job be removed if the wrong rating is used.

Different Cables, Different Choices

There are many different kinds of Category 5e or Category 6. Like any manufactured product, these can minimally meet the standard, or they may exceed the standard. Designers and end-users are urged to obtain the test data for all cables that might be considered and to compare them.

Belden makes four kinds of Category 5e and four kinds of Category 6, and each of those four types is available in plenum and riser fire ratings. Just within Belden, this gives you 16 choices of cable, a bewildering selection.

The "e" in Category 5e means "enhanced". It's an enhanced Category 5. What is enhanced is the set of parameters and tests that this cable must pass. These new tests allow this cable to run 100Base-T. In that application, all four pairs are running and the signal is divided into four parts. Not only that, but the signals run in both direction simultaneously ("duplex"), just like a telephone where you can speak and listen on the same two wires.

Table 4 shows a list of these cables and how they differ generally. Guaranteed performance specs for any cable should be available from any manufacturer. For Belden, these can be found in the Belden catalog, or even more detailed specifications at www.belden.com .

Belden Category 5e Cables
Part # Rating Gage Pairs Bandwidth Notes
1583A CMR 24 AWG Unbonded 100 MHz DataTwist 5e
1585A CMP Unbonded DataTwist 5e
1500A CMR Unbonded DataTwist 5e+
1501A CMP Unbonded DataTwist 5e+
1700A CMR Bonded DataTwist 350
1701A CMP Bonded DataTwist 350
7988R CMR Bonded Skew 9nsec/100m
7988P CMP Bonded Skew 9nsec/100m

Table 4

While Cat 5e meets the minimum requirements for 100Base-T, it became apparent that a much better cable design was needed for really good 100Base-T performance. This was Category 6. Belden Cat 6 products are listed in Table 5.

Belden Category 6 Cables
Part # Rating Gage Pairs Bandwidth Notes
7881A CMR 23 AWG Unbonded 250 MHz DataTwist 6
7882A CMP Unbonded DataTwist 6
1872A CMR Bonded MediaTwist
1874A CMP Bonded MediaTwist
7851A CMR Bonded 600e
7852A CMP Bonded 600e
7989P CMR Bonded Skew 10nsec/100m
7989R CMP Bonded Skew 10nsec/100m

Table 5

Note that some Cat 5e and some Cat 6 cables have ultra-low skew. Those low-skew numbers allow you to use these cables for RGB and VGA (and other analog and digital applications that benefit from precision multi-pair delivery) as well as for 100Base-T or 100Base-T data applications.

Color Me Fast

Most UTP data cable is available in a number of colors. While there is no "color standard", you might want to consider using different colors. For instance, all the Axia stuff might be one color, and your regular in-house networking another color, just so you can tell them apart. If you have a low-latency network (Layer 2 vs. Layer 3) you might want to color code that differently too. One broadcaster had every data installer use a different color so he could tell which installer put in which cable. Pretty clever.

Connectors And Connections

One of the most critical parts to a data network is the connections. And I do mean "connections", not "connectors", because there are two ways to make connections between cables.

The first is with a punch block, commonly called a 110-block. This is a 100-ohm, low capacitance, high quality means of connecting cables. The punch points are gas tight, so connections last a long time. This is the highest performance way of connecting category data cables together.

However, punch blocks are permanent. It is difficult to remove and reconnect cables. To disconnect and re-connect, you really need a connector. The second type of connection is a connector. The connector of choice for Category cables is called an RJ-45. This connector is very simple and fast to install.

Most data installations put jacks at the end of the installed cable, such as a plate on a wall. Then patch cables are bought pre-made to connect from the wall to the equipment. Just be aware that this point is probably the most critical for good network performance. More network failures happen here than all other places combined.

To start with, the jack must be the equal of the cable. If you put a 5e jack on Category 6 cable, you will get 5e performance. Be sure and put a Cat 6 jack on Cat 6 cable. If you can get data from the connector manufacturer, you should be able to choose the best. Belden now makes connectors for Category cables (Belden IBDN) that are very high quality and highly tested. There are other excellent brands around as well.

If you buy pre-made patch cable be aware that stranded conductor patch cables are inherently worse than the solid-conductor backbone cables. Therefore, the fewer patch cables, the shorter they are, and the higher the quality of their assembly, the better your network will run.

Be sure that your patch cables are the same Category (5e or 6) as your network. If they come with test data, or a warranty, so much the better.

Analog Applications For Data Cables

About ten years ago, with the advent of Belden MediaTwist, it became apparent that Category data cables were getting so good that they might be suitable for some non-data applications, such as analog or digital audio.

Most data cables are never tested below 1 MHz (in some cases 772 kHz). This is way above analog audio, so the actual performance of audio was not known. Figure 12 was the first test to look at the analog audio performance of premise data cable. Figure 12 shows the crosstalk performance of all four pairs averaged together, so you see the "worst case". The cable chosen was Belden 1752A, Category 5e stranded patch cable. Patch cable is possibly the worst cable made for data. But look at the results in Figure 12.

Figure 12

The worst case crosstalk is -95 dB at around 40 KHz. In the standard audio frequency band (to 20 kHz) the average of all pairs is typically -100 dB. Compares this to a CD that, when it goes "quiet" drops to maybe -90 dB, and you have to wonder why we put shields on cables. In truth foil shields are RF (high frequency) shields. They do virtually nothing at audio frequencies, and seriously nothing below 1,000 Hz. Only the twisting of the pair (and the CMRR of a balanced line).

Some observant viewers noticed that Figure 12 was FEXT (far-end crosstalk) where the signal is the weakest. Perhaps crosstalk is terrible at the other end (NEXT, "near-end crosstalk") where the signals are strongest, as in Figure 13.

Figure 13

You can easily compare Figure 12 and 13 and see that the numbers are even better in Figure 13. And what do we see? Typically, -100 dB of crosstalk averaged from all four pairs. Worst case NEXT is 45 kHz, way beyond human hearing, where the crosstalk is slightly better than -95 dB.

So we tested 1872A MediaTwist (now Cat 6). Unfortunately, I have no charts or graphs to show you because they couldn't measure it. The crosstalk was below the noise floor of the $60,000 Agilent network analyzer (-110 dB).

For digital audio, it's even easier. Digital signals are naturally noise resistant. The sampling frequency used in Axia (48 kHz) is very common. Digital audio channels on video machine are 48 kHz sampling. But that is not the frequency running on the cable. To determine that, according to the AES addendum, we must multiply that by 128. So the actual bandwidth of a two-channel digital audio bit stream at 48 kHz sampling is 6.144 MHz. We use 6 MHz as a simple frequency to test our data cable.

What is the crosstalk of Cat 5e at 6 MHz? More than -50 dB for Cat 5e, and -60 dB for Cat 6. The amazing thing is that digital signals are inherently noise-resistant. (It's very easy to tell a square wave from noise.) You only need a few dB to tell one from the other. Cat 5e and 6 give you thousands of times more crosstalk protection than you actually need.

These category cables work great for analog and digital audio, as long as they are run as a balanced line, and you can also run them as 100Base-T or 100Base-T. You can even use them to wire up a telephone!

So What Do I Choose?

Now you are well-armed to choose a "category" cable. You understand many of the considerations in design and manufacturing these cables. Besides these, you have many other factors to influence your decision. Here is a list:

  • Availability. (If you can't buy it, it doesn't matter how good it is.)
  • Consistency. (Is the roll from last year identical to next year?
  • Quality.
  • Price.
  • Ease of installation.
  • Performance.
  • Performance after installation.
  • Company track record/history.
  • Your familiarity with manufacturer and other products.
  • Recommendations from others.

If you take this list and apply it to any particular cable from a particular manufacturer, give a + for each point that cable meets, it should be very easy to judge which cable is the best for your installation. You want a cable with as many +'s as possible. A + on price alone could easily be the hardest to install with the worst performance. You are now loaded with enough questions to impress any manufacturer.

Good luck with your Axia install!



Return To